On Fri, Aug 14, 2020 at 8:34 AM Nicolás Alvarez <nicolas.alva...@gmail.com> wrote: > > El jue., 13 de ago. de 2020 a la(s) 15:49, Nate Graham (n...@kde.org) > escribió: > > On 8/13/20 12:19 PM, Ben Cooksley wrote: > > > The only way to implement what you are proposing would be to have all > > > Bugzilla actions from commit hooks take place as a 'Bot' user. > > > > That seems reasonable in the abstract, but this would mess with people's > > commit stats since their own Bugzilla accounts wouldn't get the credit > > for fixing bugs. :) > > > > So here's an idea: close the bug with the committer's own Bugzilla > > account when an account is found that matches the email address in the > > commit. Otherwise, use a bot account so that the bug at least gets > > closed as intended. Could that work? > > I think it could work, but it's not that easy to implement :) > Currently the hook code blindly sends an email to > bug-cont...@bugs.kde.org, it's all asynchronous. It doesn't know if an > account exists, or what Bugzilla does with that email once received. > To know if an account exists, we would need to make it use the > Bugzilla API...
This would make the commit hooks dependent on the availability of Bugzilla, which isn't ideal as it would mean any downtime for bugs.kde.org means downtime for invent.kde.org as well. For this reason, the commit hooks currently have no external service dependencies aside from services running on the same machine (with everything being sent asynchronously) If this were to be implemented it would need to be in the email_in.pl script (which can be found in websites/bugs-kde-org, and is a component shipped with Bugzilla), or alternatively as a Procmail filter script (which is how Activity Filter works in part) > > -- > Nicolás Cheers, Ben
